Thin Solid Films 515 (2007) 3638 3643 www.elsevier.

com/locate/tsf

Adaptive nanocomposite coatings with a titanium nitride diffusion barrier mask for high-temperature tribological applications
C. Muratore a,, J.J. Hu b , A.A. Voevodin c
a b

UTC Inc., Air Force Research Laboratory, Materials and Manufacturing Directorate, Wright-Patterson Air Force Base, Ohio, 45433 USA UES Inc., Air Force Research Laboratory, Materials and Manufacturing Directorate, Wright-Patterson Air Force Base, Ohio, 45433 USA c Air Force Research Laboratory, Materials and Manufacturing Directorate, Wright-Patterson Air Force Base, Ohio, 45433 USA Received 23 December 2005; received in revised form 27 September 2006; accepted 27 September 2006 Available online 29 November 2006

Abstract Adaptive nanocomposite coatings that demonstrate low friction throughout broad ranges of temperature, wear, humidity and other environments are currently in development. One effective temperature adaptation mechanism at temperatures 500 C is diffusion of noble metal to the coating surface, providing a low shear strength interface at the friction contact supported by a hard surface underneath. To prolong the wear lifetime of chameleon coatings relying upon this mechanism for low friction, a coating architecture incorporating a diffusion barrier mask to inhibit noble metal diffusion was employed. The diffusion barrier-capped coating provided lubrication at 500 C over ten times longer than the monolithic yttria-stabilized zirconiasilvermolybdenum coating of the same thickness without failure. Characterization of coatings after heating with surface analysis and microscopy techniques revealed the mechanism for increased coating life, and allowed for measurement of silver selfdiffusion parameters in the adaptive nanocomposite material. 2006 Elsevier B.V. All rights reserved.
Keywords: Adaptive coatings; Nanocomposite; Titanium nitride; Tribology

1. Introduction Adaptive nanocomposites, also known as chameleon coatings, are a class of materials that automatically adjust surface composition and structure to minimize friction as the ambient environment changes [15]. Adaptation results from amorphous and nanocrystalline inclusions that can transform into lubricious macro-phases on the friction contact surface when exposed to changes in temperature, relative humidity, and/or wear. A series of adaptive coatings incorporating metals in a nanocrystalline/amorphous yttria-stabilized zirconia (YSZ) matrix were designed especially for use as solid lubricants from 25700 C. These coatings incorporated soft noble metals for lubrication at low and moderate temperatures, and other transition metals known to form lubricous oxides at higher temperatures. A YSZAgMo coating, which provides lubrication by forming a silver rich surface at 300500 C, and a
Corresponding author. E-mail address: chris.muratore@wpafb.af.mil (C. Muratore). 0040-6090/$ - see front matter 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2006.09.051

MoO3 surface at temperatures above 500 C, was selected for the present work. While the adaptive lubrication mechanisms are effective in reducing the friction of pure YSZ coatings from 0.8 to 0.2 [2,6], the coating lifetime is generally limited to 10,000 cycles or less. One reason for the relatively short lifetime is the homogeneous diffusion of silver or gold to the entire coating surface, illustrated in Fig. 1a and b. The coatings are produced in a metastable condition, by forced co-deposition of metal and crystalline YSZ at substrate temperatures of 150 C. Asdeposited, these nanocomposite coatings have noble metal atoms distributed in a YSZ matrix, as either dispersed atoms or nanoinclusions in distorted YSZ nanocrystals or amorphous YSZ regions. Heating activates metal diffusion out of highly strained sites in the YSZ matrix to the open surface, where metal grain nucleation and growth is thermodynamically driven by the reduction of system potential and surface energy. Upon heating, the noble metal rapidly distributes itself over the entire surface, as the coating bulk becomes depleted in lubricious metal phases. While the depletion of the metal from the composite matrix to the

C. Muratore et al. / Thin Solid Films 515 (2007) 36383643

3639

Fig. 1. Schematic of (a) the monolithic adaptive coating, and (b) its response to heating. The pinholes in the as-deposited TiNYSZAgMo coating (c) should allow for control of silver migration to the coating surface upon heating (d).

surface is essential for chameleon adaptive behavior, the controlled restriction of diffusion is needed to extend the coating service life. Transition metal nitrides are documented as effective diffusion barrier materials for noble metals [79]. It was presumed that a TiN barrier layer with a random array of pinholes could be coupled with an adaptive coating layer (Fig. 1c) to limit lubricant transport to the coating surface only through those holes (Fig. 1d), thus reducing the rate of lubricant depletion in the adaptive coating layer and prolonging the life of the coating during sliding wear at elevated temperatures. A YSZAgMo coating design incorporating a TiN surface mask layer was

Fig. 2. Schematic of the hybrid filtered vacuum arc/magnetron sputtering/pulsed laser deposition system.

Fig. 3. Scanning electron micrographs of the as-deposited coatings: (a) monolithic YSZAgMo, (b) monolithic TiN, and (c) TiN/YSZAgMo.

3640

C. Muratore et al. / Thin Solid Films 515 (2007) 36383643

arc plasma source fitted with a titanium cathode. After cleaning, the titanium ion-based plasma was used to deposit a thin (b 80 nm) adhesion layer. Pulsed laser deposition (PLD) of YSZ was then initiated while operating the arc source to produce a 300 nm graded Ti/YSZ transition layer to further promote coating adhesion. The PLD process employed a KrF excimer laser (248 nm) to deliver 800 mJ, 25 ns wide UV pulses (248 nm) at incrementally increasing pulse frequencies from 2 30 Hz to a 5 wt.% yttria-stabilized zirconia (YSZ) target, with the substrates biased to 150 V dc and maintained at 150 C. Programmable mirrors were used to direct the laser to random positions on the surface of the rotating, 5.0 cm diameter YSZ
Fig. 4. Friction traces of the three coatings in air at 500 C.

used to test this concept in the current work. The TiN/YSZ AgMo coating architecture also allowed for determination of the diffusion rates and coefficients of silver through the adaptive coating matrix to the coating surface. 2. Experimental procedure Coatings were deposited with a hybrid filtered vacuum arc/ magnetron sputtering/pulsed laser deposition technique [10,11] in a stainless steel chamber shown schematically in Fig. 2. For each deposition, six 1 in. diameter M50 steel substrates were degreased ultrasonically and loaded into the chamber. The chamber was pumped with a 520 l s 1 turbomolecular pump to a base pressure of 7 10 6 Pa or less. The substrates were heated to 150 C, then bombarded with energetic argon ions (800 eV) and metal ions ( 650 eV) to clean the surface before deposition. The metal ions were produced with a filtered

Fig. 5. Scanning electron micrograph of the TiN/YSZAgMo wear track after wear testing for 50,000 cycles at 500 C.

Fig. 6. Scanning electron micrographs of (a) the TiN/YSZAgMo coating surface and (b) the surface of the monolithic adaptive coating after wear testing at 500 C.

C. Muratore et al. / Thin Solid Films 515 (2007) 36383643

3641

target. The chamber was then filled with 1.6 Pa Ar (99.99% pure) at a flow rate of 100 sccm. The arc source was turned off and two magnetrons, each fitted with a 3.3 cm pure metal target and configured as shown in Fig. 2, were used to add silver and molybdenum to the YSZ. After depositing 1.1 m of adaptive YSZAgMo nanocomposite material, a 300 nm TiN layer was deposited by flowing 30 sccm nitrogen into the chamber while operating the filtered titanium arc plasma source. For this final step, the uppermost magnetic field coil on the filtered arc source was deliberately defocused to reduce the ion flux density and energy to produce a TiN layer with the desired array of randomly dispersed holes. Monolithic YSZAgMo and TiN coatings were also produced under the same applicable conditions as those described above for comparison in sliding friction. X-ray photoelectron spectroscopy (XPS) was used to determine that the YSZAgMo coatings contained 24 at.% Ag and 10 at.% Mo. XPS was also used to determine that the TiN coatings were stoichiometric. The as-deposited coatings were examined with electron microscopy techniques, and then subjected to ball-on-disc wear testing at 500 C in air with 40% relative humidity. A 6.35 mm silicon nitride ball with a 1 N load was used for all tests. The sliding rate was 0.075 m s 1, with the disk rotating at approximately 75 rpm. The tribometer thermocouple was calibrated before testing with a separate thermocouple spot-welded to an uncoated M50 substrate that was heated without rotation. Wear tests were started after the coating samples were heated to the desired temperature ( 20 min) and allowed to equilibrate for 5 min. Upon completion of the ball-on-disk tests, samples were immediately removed from the furnace and allowed to cool in air. The coatings were examined again with electron microscopy and X-ray energy dispersive spectroscopy, after wear testing. A series of annealing experiments were also performed to better understand the adaptive lubrication mechanisms demonstrated by the TiN/YSZAgMo coating. Three different samples were annealed for 45 min at 425, 500 and 575 C. A

focused ion beam (FIB) along with scanning and transmission electron microscopes (SEM and TEM) were used to produce and examine cross sections of the coatings after heating [6,12]. 3. Results Fig. 3ac are scanning electron micrographs of the asdeposited coatings. In Fig. 3a, the surface of the adaptive YSZ AgMo coating is shown to have sub-micron clusters of silver and larger YSZMo grains. X-ray diffraction and crosssectional transmission electron microscopy revealed that the coating consisted of an amorphous/nanocrystalline YSZ matrix with nanocrystalline metal inclusions [10]. Fig. 3b shows the monolithic TiN coating with randomly dispersed holes on the surface. Fig. 3c shows the TiN/YSZAgMo coating. The YSZAgMo coating can be observed through the TiN holes, which looks similar to that shown in Fig. 3a. Fig. 4 shows the friction behavior of each film in the ball-ondisc tests at 500 C. The monolithic adaptive coating demonstrated a steady state friction coefficient of 0.4 before failure at 4500 cycles, as indicated by a sudden increase in the friction coefficient. The friction coefficient of the titanium nitride was maintained at 0.5 for approximately 12,000 cycles before increasing to 0.65, which is consistent with the friction coefficient observed for the uncoated M50 substrate at 500 C under similar conditions [6]. The TiN/YSZAgMo coating did not fail over the 50,000 cycles, and maintained a friction coefficient of 0.4. Micrographs of the TiN/YSZAgMo wear track are shown in Fig. 5. X-ray energy dispersive spectroscopy measurements were used to find that the wear track was composed of a layer of silver smeared over the TiN. A micrograph of the heated TiN/YSZAgMo surface away from the wear track is shown in Fig. 6a. From the higher-magnification inset, it appears that silver diffused out of the lower YSZAgMo layer through the holes in the TiN masking layer, whereas in the monolithic YSZAgMo coating, the equiaxed grains of silver were homogeneously distributed over the entire coating surface

Fig. 7. Scanning electron micrographs showing (a) the surface of the annealed TiN/YSZAgMo coating after 45 min. The rectangle indicates the cross-sectioned area in (b). An XEDS map of a similar cross-section (c) shows that the silver diffused laterally to the coating surface, out of the pinhole, leaving a clearly defined region depleted in silver.

3642

C. Muratore et al. / Thin Solid Films 515 (2007) 36383643

Fig. 8. Scanning electron micrographs of the TiNYSZAgMo coatings after annealing for 45 min at (a) 425 C, (b) 500 C, and (c) 575 C.

(Fig. 6b). Several samples were subjected to a 12 h duration wear tests, with a subsequent analysis by both Auger and XPS techniques. This analysis confirmed that the surface of the adaptive composite coating which was capped with a holey TiN masking layer consisted of silver with some oxygen. Measurement of the wear scar to determine the wear rate was inconclusive since the total thickness of the coating increased by an unknown amount because of silver segregation to the surface, as shown in Fig. 6a. The relatively low wear rate can be predicted however, from the fact that XPS in the wear scar of the TiN/YSZAgMo coating did not reveal any substrate elements, only silver, titanium, nitrogen and oxygen. Thus, 50,000 sliding cycles (or 3 km) were not enough to wear through the total coating thickness. Another TiN/YSZAgMo sample was heated to 500 C in the tribometer furnace and maintained at that temperature for 45 min to observe evolution of the surface morphology and composition in the early stages of heating. Fig. 7a shows the surface of the coating after annealing. The FIB was used to cut out sections across the diameters of silver islands, as shown in Fig. 7a and b. EDS composition maps were generated for all of the elements detected in the cross sections. Silver EDS maps revealed that silver migrated upwards, from below the TiN layer to the surface, leaving behind a section of YSZMo coating material (Fig. 7c). From the micrograph in Fig. 7c, it was determined that a silver atom had to travel an average minimum distance of 6 m from the border of the silver depleted region inside YSZAg Mo layer indicated on the figure. Based on this observation of the diffusion controlled depletion length, the diffusivity (D) of silver in YSZ at 500 C was estimated to be 1 10 10 cm2 s 1 assuming parabolic diffusion, where x (Dt)1/2 . It was also observed that a 17 m average silver island size corresponded to an average depletion length of 6 m. As-deposited samples of the TiN/YSZAgMo coating were annealed at 425 and 575 C for 45 min to observe the temperature dependence on diffusion length. This somewhat narrow range of annealing temperatures was selected to minimize the difference in

heating and cooling time. The time to heat to 425 C was approximately 15 min compared to 20 min to heat to 575 C. The diffusion distance for the samples annealed at 425 and 575 C were obtained by assuming a linear relationship between the silver island diameter and the length of the depleted silver zone beneath the TiN layer for each sample, proportional to that observed for the sample annealed at 500 C. Electron micrographs of the surfaces of the coatings annealed at the three different temperatures are shown in Fig. 8. The average island diameters were 6 mm for the sample heated to 425 C (Fig. 8a), and 26.5 mm for the sample heated to 575 C (Fig. 8c). A log plot of the silver migration rate squared versus inverse temperature yielded a straight line shown in Fig. 9. The migration rate was based on an average diameter of at least 10 Ag islands located at least 2 island diameters away from neighboring islands for each temperature. The data yielded an estimated diffusion prefactor (Do) of 4.0 10 4 cm2 s 1 and activation energy of 1.0 eV/atom for diffusion of silver in the YSZ-based nanocomposite material.

Fig. 9. Arrhenius plot of the temperature dependent migration rate data. The line is a least squared fit to the data.

C. Muratore et al. / Thin Solid Films 515 (2007) 36383643

3643

4. Discussion Noble metal diffusion from the YSZ-based matrix to the composite surface at temperatures 300 C has been observed in several other works [2,4,6,12], and similar phenomena have also been reported in other composite materials [13,14]. The migration of silver to the surface at elevated temperatures is of interest for tribological applications, as silver demonstrates a relatively low shear strength, and can reduce the friction coefficient on hard surfaces during sliding wear [15,16]. The monolithic YSZ20Ag10Mo coating in this work demonstrated a friction coefficient of 0.4 at 500 C, which has been attributed to lubrication of the hard YSZ by silver [6]. The lifetime of the monolithic coating was short compared to the TiN-capped adaptive coating, because nearly all of the silver diffused to the surface in approximately 45 min. Comparison of the surface shown in Fig. 7a to that in Fig. 6b reveals that the rate of Ag depletion in the TiN/YSZAgMo coating was substantially slower, as both samples were maintained at 500 C for the same amount of time. While the rate of silver transport was reduced in the TiN capped coating, Fig. 4 shows that the friction coefficient was comparable to that observed in the monolithic coating with rapid silver transport and complete surface coverage. The micrographs of the wear track in Fig. 5 indicate that silver was distributed over the TiN surface, which showed little wear after the 50,000 cycle wear test at 500 C. Nevertheless, 0.4 appears to be the lower limit of friction coefficient for hard surfaces lubricated by diffused silver. Adaptive composite materials exhibiting lower friction coefficients over a broad temperature range demonstrate similar diffusion mechanisms [17], and will be used in similar, future studies. The Arrhenius plot shown in Fig. 9 suggests that diffusion through pinholes in the diffusion barrier layer is the means by which silver is supplied to the coating surface during heating. Although the estimated diffusion prefactor and activation energy were low compared to typical published values in the literature for silver volume self-diffusion in metal [18] or ceramic matrices [19], the values were consistent with published values for silver grain boundary self-diffusion [20,21] and for self-diffusion in amorphous materials [22]. Indeed, the relationship of the diffusion prefactor to the activation energy was consistent with that discussed by Naundorf et al. for selfdiffusion of metals in an amorphous matrix [22]. Moreover, the relatively large surface and strain energies of the silver inclusions, immiscible in the YSZ matrix, may have further contributed to rapid diffusion and coalescence of silver to the coating surface [13]. 5. Conclusions An adaptive solid lubricant coating layer capped with a holey TiN layer demonstrated a lower friction coefficient and wear rate than monolithic TiN or adaptive coatings. The in-

creased wear life was attributed to the reduced depletion rate of silver lubricant resulting from restricted flow through the holes in the TiN. The friction coefficient was typical of soft noble metal lubrication of hard surfaces. The mechanism of silver transport to the coating surface during heating was identified to be diffusion from the adaptive composite matrix to the free surface of the coating. The architecture of the coating allowed for a measurement of the diffusion prefactor (Do) and activation energy for silver in nanocrystalline/amorphous YSZ, which were 4.0 10 4 cm2 s 1 and 1.0 eV/atom, respectively. These values were attributed to the large number of defects in the YSZ matrix in addition to the reduction in surface and strain energy accompanied by migration to the surface. Acknowledgements The Air Force Office of Scientific Research is gratefully acknowledged for funding this research. The authors also wish to thank John E. Bultman and Art Safriet for their technical assistance. References
[1] A.A. Voevodin, J.S. Zabinski, Thin Solid Films 370 (2000) 223. [2] A.A. Voevodin, J.J. Hu, T.A. Fitz, J.S. Zabinski, Surf. Coat. Technol. 146 147 (2001) 351. [3] A.A. Voevodin, J.G. Jones, J.J. Hu, T.A. Fitz, J.S. Zabinski, Thin Solid Films 401 (2001) 187. [4] A.A. Voevodin, J.J. Hu, T.A. Fitz, J.S. Zabinski, J. Vac. Sci. Technol., A, Vac. Surf. Films 20 (2002) 1434. [5] A.A. Voevodin, J.S. Zabinski, Compos. Sci. Technol. 65 (2005) 741. [6] C. Muratore, A.A. Voevodin, J.J. Hu, J.S. Zabinski, Wear 261 (2006) 797. [7] D. Adams, G.F. Malgas, N.D. Theodore, R. Gregory, H.C. Kim, E. Misra, T.L. Alford, J.W. Mayer, J. Vac. Sci. Technol., B 22 (2004) 2345. [8] L. Chen, Z. Yuxiao, P. Nyugen, T.L. Alford, Mater. Chem. Phys. 76 (2002) 224. [9] M.-A. Nicolet, Thin Solid Films 52 (1978) 415. [10] C. Muratore, A.A. Voevodin, J.J. Hu, J.G. Jones, J.S. Zabinski, Surf. Coat. Technol. 200 (2005) 549. [11] A.A. Voevodin, M.A. Capano, A.J. Safriet, M.S. Donley, J.S. Zabinski, Appl. Phys. Lett. 69 (1996) 188. [12] C. Muratore, A.A. Voevodin, J.J. Hu, J.S. Zabinski, Tribol. Lett., in press. [13] C.W. Bates, Q.Y. Chen, Mater. Lett. 23 (1995) 7. [14] J.L. Endrino, J.J. Nainaparampil, J.E. Krzanowski, Surf. Coat. Technol. 157 (2002) 95. [15] F.P. Bowden, D. Tabor, J. Appl. Phys. 14 (1943) 141. [16] A. Erdemir, R.A. Erck, G.R. Fenske, H. Hong, Wear 203204 (1997) 588. [17] C. Muratore, A.A. Voevodin, J.J. Hu, J.S. Zabinski, World Tribology Congress, Washington DC, U.S.A., September 1216 2005, ASME Proceedings of the Third World Tribology Congress, 2005, p. 5. [18] H. Gleiter, Prog. Mater. Sci. 33 (1990) 223. [19] O. Gorur, C. Terzioglu, A. Varilci, M. Altunbas, Supercond. Sci. Technol. 18 (2005) 1233. [20] S. Schumacher, R. Birringer, R. Strauss, H. Gleiter, Acta Metall. 37 (1989) 2485. [21] J. Sommer, C. Herzig, J. Appl. Phys. 72 (1992) 2758. [22] V. Naundorf, M.-P. Macht, A. Bakai, N. Lazarev, J. Non-Cryst. Solids 224 (1998) 122.